Edward N. Lorenz, a theoretical meteorologist at MIT, published 1963 the first simulation of a chaotic system in the Journal of Atmospheric Sciences (Vol.20, p. 130ff.) under the title "Deterministic nonperiodic flow". He called the strange sensitivity of his system to small variations in the initial conditions "Butterfly Effect". This term was introduced to illustrate this sensitivity or "Error Growth" in the context of weather forecast: If the disturbance of a single butterfly is not taken into account in simulations of the atmosphere, the calculated forecast will be wrong after a certain time. Because of the impossibility to include detailed calculations of small scale turbulence in computer simulations of the atmosphere, prediction is limited by principle.
In the 1960s, limitation of predictability in a deterministic system was unthinkable and so, a real system was needed to demonstrate that chaos and the butterfly effect were realities and not mere mathematical artefacts (we call the three deterministic and nonlinear normal differential equations "Lorenz Equations"). W.V.R. Malkus, a mathematician at MIT, realized that the Lorenz-Equations can be transformed into the equations of motion of a waterwheel. This waterwheel was built at MIT in the 1970s and helped to convince the sceptical physicists of the reality of chaos. For the equivalence of the waterwheel with the Lorenz System see e.g. Steven H. Strogatz, Nonlinear Dynamics and Chaos, Addison-Wesley, 1994, p. 301-311.
The waterwheel at the Fachhochschule in Brugg-Windisch (Switzerland) is not exactly identical with the Lorenz Equations, because the water leakage rate is not exatly proportional to the water content in the respective bucket.
back to Waterwheel Lab